With TSMC warning that Nvidia’s H100 AI GPUs will continue to see short supply over the next 18 months due to inadequate chip-on-wafer-on-substrate (CoWoS) packaging capabilities, other companies see an opportunity to step in and make a compelling case. The demand for generative AI hardware. One such company is a new startup called d-Matrix, which recently managed to secure US$110 million from investors like Microsoft by demonstrating the incredible speed of its Corsair C8 compute card that has 9x the generative throughput of the Nvidia H100 GPU. AI workloads.
d-Matrix’s Digital-in-Memory Computer (DIMC) technology achieves impressive speeds. Unlike Nvidia’s A100 and H100 GPUs that are designed for a wide variety of high-performance computing tasks, d-Matrix’s Corsair C8 cards are specifically designed for generative AI and inference workloads. A single Corsair C8 card combines 2,048 DIMC cores grouped into 8 chiplets with a total of 130 billion transistors and also features 256 GB of LPDDR5 RAM to easily fit full AI models into memory. All chiplets are interconnected via PCIe 5.0 lanes with 1 TB/s bandwidth resulting in a total die-to-die bandwidth of 8 TB/s. d-Matrix also includes 2 GB SRAM with 150 TB/s bandwidth. Depending on the AI workload, the Corsair C8 delivers between 2400 and 9600 TFLOPS.
Thanks to optimized scaling capabilities and the frictionless nature of the open source software stack, d-Matrix claims that a cluster of Corsair C8 cards can achieve 20x better power efficiency, 20x lower latency and up to 30x lower costs than a similar solution from Nvidia. .
Buy Lumina 4K Webcam powered by AI on Amazon
I first stepped into the wonderful IT&C world when I was about seven years old. I was instantly fascinated by computer graphics, whether it was for games or 3D applications like 3D Max. I am also an avid science fiction reader, astrophysics enthusiast and crypto geek. I started writing PC-related articles for Softpedia and some blogs in 2006. I joined the Notebookcheck team in the summer of 2017 and am currently a senior tech writer covering processor, GPU and laptop news.